Authoring features respecting the constraints of time is hard, but required, when computing from event-based data directly. We'll review the limitations of data science tools and deep dive on how we've solved the problem. Attendees will gain understanding of temporal processing to create and operate predictive models with event-based data.

Talk date and time: Saturday 9/18/21 4:00 PM - 4:40 PM Pacific

Talk Description

Come join our VP of Product Charna Parkey and CEO, Co-Founder Davor Bonaci discussing how to author features respecting the constraints of time. Feature engineering is supposed to be an iterative process, transforming raw data into training examples and feature vectors. Iteration is key -- but, each cycle should include trying new ideas offline, as well as testing in production.

Offline experimentation requires historical event-based data to compute training examples at the right points-in-time -- quickly, without waiting for complex pipelines to be built just to determine if a feature will be useful. Then, in the latter part of each iteration cycle, we need to test the new model live -- without worrying about offline and online discrepancies.

Feature stores are the newest idea that is supposed to help us, but it turns out that’s not enough. In this session, you’ll learn how to craft production-ready features and build training datasets at the right points-in-time from event-based data. Specifically, we’ll be covering strategies for powering feature stores with a feature engine to:

  • Compute directly from event-based data to try new features

  • Iterate on feature definitions and time selection across historical data instantly

  • Join values between different entities at precise times — without leakage

  • Eliminate data discrepancies in production

Come join us to learn how to finally iterate on amazing ML models with event-based data.

Data Con LA 2021 Details

The Largest Data Conference in Southern California.

Spearheaded by Subash D’Souza and organized and supported by a community of volunteers, sponsors and speakers, Data Con LA features the most vibrant gathering of data and technology enthusiasts in Los Angeles.

Data Con LA began as Big Data Day LA in 2013, with just over 250 attendees. We have since grown to over 550 attendees in 2014, 950+ attendees in 2015, 1200+ attendees in 2016, and 1550+ attendees in 2017. In 2018, we re-branded ourselves from Big Data Day LA to Data Con LA with over 1800 attendees and over 2000 in 2019. In response to the COVID-19 pandemic, DCLA had its first successful virtual conference in 2020 with over 1000 virtual attendees.

Register Now